The marketing of web services on the internet continues to increase, resulting in an increasing number of web services and, therefore, operations offering equivalent functionalities. As a consequence, finding an appropriate web service (operation) for a particular task has become a difficult challenge, taking a lot of time and leading to an insufficient selection of relevant services. This work aims to propose a new approach facilitating the search and localization of relevant web services (operations) in an acceptable time while ensuring the totality of the response. This approach is divided into three crucial phases. The first step involves collecting web services from various universal description, discovery, and integration (UDDI) registries and different domains and forming specialized sub-registries. The second phase involves the extraction of operations from various services, followed by a similarity study whose goal is the formation of clusters of similar operations. The third phase processes user requests by identifying the desired features. A list of operations is then provided to the client, including the non-functional properties, from which they select the one that best meets their needs and begin to invoke it.
Read full abstract